Catheter ablation techniques to treat chronic reentrant atrial fibrillation (AF) have been largely unsuccessful. We postulate that a major determinant of failure is the extreme variability in atrial structure among patients, so that no single catheter ablation strategy will be successful; rather, therapy must be individualized for a given patient. We hypothesize that incorporating specific anatomic information about an individual patient's atrial structure into the ablation strategy will increase the likelihood of success. In this project, we propose to perform the fundamental research necessary to bring patient-specific ablation therapy to the animal laboratory and, ultimately, to human patients. Using three-dimensional endocardial mapping techniques during electrophysiologic diagnostic studies, patient-specific atrial anatomy will be reconstructed from data sets containing the locations and electrogram characteristics of atrial points obtained by a catheter probe. Simulations of AF will be performed using the reconstructed atria. Within this framework, different ablation strategies will be tested in silico to determine how ablation lesions can be successfully performed for each specific individual anatomic structure. This work is designed to lead directly to testable ablation strategies and to novel clinical paradigms in the treatment of AF.